One of the bizarre things about the patent system is the "presumption of validity," in which a patent officially has to be presumed valid
. Conceptually, this makes very little sense. Patents grant a pretty broad monopoly on "inventions" for an extended period of time... based entirely on approximately 18 hours that a patent examiner has to spend looking over the thing. Do we really think that a patent examiner gets things right most of the time? It seems that even the US Patent Office's own data
shows that's simply not true. A friend pointed me to the USPTO's recently released data concerning re-exams
(pdf and embedded below), which demonstrates in great detail why patents shouldn't be presumed valid. Basically, the data suggests that an awful lot of patents were handled poorly.
The document notes that 92% of re-exam requests are granted -- meaning that nearly all re-examination requests lead to a re-examination by the Patent Office. So, if most patents were well constructed in the first place, you would imagine that most of them would come through the re-examination process unscathed with no changes, right? Only if patent examiners were really bad at their jobs would a large percentage of patents need to be changed or rejected completely on re-exam. Given the "presumption of validity" that grants a monopoly, and the massive dollar amounts that patents sell for and are able to extract in settlements, you'd think that re-examined patents must normally confirm the original diagnosis. Hell, given that information, I'd hope that at least
around 95% of patents, having passed the approval process, would be solid enough to survive the re-exam process untouched.
If the number was below 90%, I'd think the system was in trouble and needed some fixing. If it was below 70%, I'd think that we should be declaring the system a failure. If it was below 50%, I'd be questioning the entire basis of the patent system. So what is it?
Would you believe that only 22% of re-examined patents have all claims confirmed?
22%! That means that 78% of all patents that are granted a re-exam had serious problems
with their original claims -- and remember, 92% of re-exam requests are granted. All these patents were initially approved and enjoyed the presumption of validity, which would may have cost companies millions (or more)
. This isn't just a failing grade. This is an epic disaster
. It's true that 67% of the re-examined patents still are allowed with "claim changes", and only 11% are completely rejected, but those numbers are little comfort when we're told that we need to presume all of the claims in all patents are perfectly valid.
Now, some might claim that this number is perfectly fine, because only bad patents get re-exam requests. In fact, you could argue that perhaps these numbers show the system is working in that bad patents get re-exam requests and good patents remain valid. But there's little to no evidence to support that. Already, those who dislike patent re-exams are claiming that patent re-exams are abused
with too many good patents getting re-examined. So it certainly appears that all sorts of patents get re-examined... and a very large percentage of them appear to turn out to have been mistakenly granted.
This highlights, in pretty stark contrast, just how broken and completely arbitrary
the system is. For a system like this to be valid, it should be formalized and repeatable
. It needs to be based on objective information, not the random subjective opinions of a particular examiner. Yet the data suggests that's exactly what's happening, meaning that we're handing out hundreds of thousands of monopolies based on the mere whims of patent examiners, who haven't been shown to be even remotely consistent, and who have very little time to actually examine what it is they're granting monopolies over.
How does anyone
consider that to be a reasonable system?